Principal Component Analysis (PCA)

Principal Component Analysis (PCA) is a technique that linearly recombines all identified metabolites to form a set of new comprehensive variables. From these, 2-3 comprehensive variables are selected based on the analysis objective to reflect as much information of the original variables as possible, thereby achieving dimensionality reduction. Conducting PCA on metabolites can also reflect the variability between and within groups overall. Overall sample PCA analysis uses the PCA method to observe the overall distribution trend among all group samples, identify potential outlier samples, and determine whether to remove outliers by considering various factors (number of samples, rarity of samples, degree of dispersion). The PCA score plot for all samples is shown below (PCA score plot for pairwise sample analysis).

Figure 1 PCA Score Plot

Biotree uses XCMS software to extract metabolite ion peaks. Peaks obtained from 25 experimental samples and QC samples are normalized and then analyzed using PCA. As shown in the figure, QC samples (in black) are closely clustered together, indicating good stability of the instrument analysis system in this experiment. The test data is stable and reliable, and the metabolic profile differences obtained in the test can reflect the biological differences between samples.

Figure 2 PCA Score Plot of Total Samples